ICAR keynote presentations Genomic selection and its ... · 9. Juni 2010 Seite 1 ICAR keynote...
Transcript of ICAR keynote presentations Genomic selection and its ... · 9. Juni 2010 Seite 1 ICAR keynote...
9. Juni 2010 Seite 1
ICAR keynote presentations
Genomic selection and its potential to change cattle breeding
Reinhard Reents
- Chairman of the Steering Committee of Interbull- Secretary of ICAR- General Manager of vit, IT solutions for animal production
N:zws/folien/rreents/genetic_evaluation_riga2010_reents.ppt
9. Juni 2010 Seite 2
Outline
Where we come from
New opportunities due to genomics
Challenges for breeding programs
Outlook
9. Juni 2010 Seite 3
Selection from about 1960 to ~ 2009
Quantitative-genetic concepts(Wright, Lush, Henderson) -> additiv genetic model
Genetic evaluationSeparate phenotypic observations (eg 9850 kg milk) in
• additiv genetic effect estimated breeding value (eg. + 1430 kg M)• Systematic environmental effect• Residual effect
Ranking based on estimated breeding values (EBVs)
9. Juni 2010 Seite 4
A large proportion of genetic gain is achieved by a very high selection intensityon the male side (one bull returning to service out of ~10-15 test bulls
with 60.000 to 100.000 doses of semen per proven bull and year)
Necessary elements for this ‚traditional breeding program‘Phenotypic observations
• Milk yield, somatic cell counts, type traits, etc. Pedigree dataData structure (across herds/environments)
Artificial insemination gives optimal structure to estimate EBVs that rank theanimals best and unbiased in many environments
• Algorithms (Henderson, Schaeffer&Kennedy, Misztal, etc) and computing power
BLUP methodology, which result in highly reliable EBVs (85-99%) for bullswith a progeny test of 100-150 daughtersTransformation of these EBVs since 15 years via Interbull MACE
Bulls that are marketed worldwide
Selection on EBVs in AI breeding programs
9. Juni 2010 Seite 5
1995 Production
1999 Production Type
Workability
2001 Production Type Cellcount
2004 Production Type Cellcount Longevity
2008 Production Type Cellcount Longevity Calving Fertility
2005 Production Type Cellcount Longevity Calving
2007 Production Type Cellcount Longevity Calving Fertility
Portfolio of Interbull evaluations
9. Juni 2010 Seite 6
Genetic trend in Holstein bulls(MACE Evaluation Interbull, Holstein AI bulls (50,000); Mean = 100, SD = 10; Data Sept. 2007, Berglund, 2008)
EBV
PROT=Protein yield Jorjani, 2008
76
82
88
94
100
106
112
1985 1990 1995 2000
PROT
Year of birth of AI bulls
9. Juni 2010 Seite 7
but
Genetic gain / costsHigh generation interval due to progeny test
Expensive, due to keeping of many waiting bulls Genetic gain per year not very high
Reasons for this: Although one can select young bulls for progeny test on very reliable EBVs
• (bull dams ~60% r² and bull sires ~95% r²)reliability of a pedigree index =0,5 EBV sire + 0.5 EBV dam + MSis as low as 25-35%
• MS = Mendelian sampling in the Meiosis
9. Juni 2010 Seite 8
Meiosis (simplified)
Recombination of chromosome fragments during Meiosis (I)Random allocation of the chromosomes, that were inherited from sire and dam, to the gametes
Large number of different zygotes (offspring) from the same pair of parents Superior sires could only be identified by generating and testing ~ 100 daughters
9. Juni 2010 Seite 9
New opportunities: SNP - genotyping
SNP = Single-nucleotide polymorphismGenotype = Which alleles of the nucleotides A-T,C-G an animal carriesGenome = contains 3 billion base pairsCa. 50.000 SNPs at a cost of about 200 EUR
Anim. n:...AGGCACC GCAATCCACG GAGGCAACGC CCTCACCGGA GGTTTCGCTC TCCACGG... ...AGGCACC GCAATCCACG GAGGCAACGC CCTCACCGGA GGTTTCGCTC TCCACGG...
Anim. 3:...AGGCACC GCAATCCACG GAGGCTACGC CCTCACCGGA GGTTTCGCTC TCCACGG... ...AGGCACC GCAATCCACG GAGGCAACGC CCTCACCGGA GGTTTCGCTC TCCACGG...
Anim. 2: ...AGGCACC GCAATCCACG GAGGCAACGC CCTCACCGGA GGTTTCGCTC TCCACGG......AGGCACC GCAATCCACG GAGGCAACGC CCTCACCGGA GGTTTCGCTC TCCACGG...
Anim. 1: ...AGGCACC GCAATCCACG GAGGCTACGC CCTCACCGGA GGTTTCGCTC TCCACGG......AGGCACC GCAATCCACG GAGGCTACGC CCTCACCGGA GGTTTCGCTC TCCACGG...
Genotype:
TT
AA
AT
AA
Eg position on chromosome 6 # 43.675.239
9. Juni 2010 Seite 10
Use in practical application
Lab is an important part, BUT SNP data has per se no information on ‚traits‘
Steps (eg Meuwissen et al 2001):
Genotype animals that have reliable EBVs from ‚conventional‘ genetic evaluation Calculate regression formulas from a ‚reference population‘ so that SNPs explain well the conventional EBVUse the regression formulas derived by historic data to evaluate young animalsSelect among these young animals
9. Juni 2010 Seite 11
Need for collaboration US / CAN Eurogenomics (Hol: DEU, FRA, NLD, DFS)Intergenomics (BS: AUT, CHE, DEU, FRA, ITA, SLO, USA )
Key factor in application: Size of reference population
Goddard
9. Juni 2010 Seite 12
Genomic EBVs
Statistical models work well
Difficult to have same ranking for cows and bulls with only phenotypic data Genomic plus phenotypic dataImportance of Interbull validation procedure
Implemented in various countriesNeed for converting GEBVs across countriesInterbull activities
• GMACE
Intense use of GEBVs can significantly increase genetic gainGenetic gain doubled (eg Schaeffer 2006)
9. Juni 2010 Seite 13
Increase in reliability of the BVs for AI bulls
Al Bull Milk Yield Al Bull longevity
relia
bilit
y(r
²)
relia
bilit
y(r
²)
Age (years)
GEBVDGVEBV First corp.daug.Second corp. daugt.
GEBVDGVEBV First corp.daug.Second corp. daugt.
Age (years)
9. Juni 2010 Seite 14
Increase in reliability of the BVs: cows
Cow Milk Yield Cow Longevity
Age (years) Age (years)
relia
bilit
y(r
²)
9. Juni 2010 Seite 15
Practical implementation in breeding programs
GEBVs allow to select on the Mendelian sampling term of young animals (without own phenotypes or progeny with phenotypes)GEBV = 0,5 EBV sire + 0.5 EBV dam + MS
This is very good approximated by Σ SNP effects
Strategy old:Identify superior parents
mate them progeny test (with 100 daughters) the highest young candidates
Select among bulls with highest EBVs from progeny testWide use of these bulls
Success of a breeding program was closely linked to population size (test population) of cows under milk recording
9. Juni 2010 Seite 16
Practical implementation in breeding programs
Strategy new:Identify superior animals (r² ~ 75%)
mate them progeny test (100 daughters) the highest young candidates
Select among bulls with highest EBVs from progeny testWide use of these bulls ??
Success of a breeding program is closely linked to Genomic evaluation system ability to calculate GEBVs with high r²Sourcing of promising candidates to calculate GEBVs on themOptimised breeding program to have the
• Optimal number of candidates for genotyping • Optimal selection intensity on them• Number of insemintations per young bull • Etc. (see presentation of Meuwissen in joint ICAR / IB session on genomics)
Success of breeding program is not directly linked to population under milk control any more• Entry for new players in the field
9. Juni 2010 Seite 17
Challenges for breeding programs and herd recording
Rapid developments cause large challenges for cooperative breeding programsTechnological issuesInteraction with the breeders
• As providers of superior genetics eg for AI centres• As users of semen
Availability of phenotypic dataStandard traits:
Necessary for updating the genomic prediction formulaNot directly needed for individual animals breeding value (different to old situation)
New traitsFunctional traits (eg as defined by the ICAR WG ‚Functional Traits‘)New traits e.g. composition of milkhave to be collected on a large number of animals (mostly cows) to first calculate EBVs for these traits and derive prediction formulas for GEBVs
Commercial companies may try to take over part of the cattle breeding industryIssues
• Technology• Patents• Data availability
9. Juni 2010 Seite 18
Areas for future research
Optimal statistical modelStructure and size of reference population
• Incl. Pooling of reference samples across countriesBias due to preselection of bulls on GEBVs in conventional genetic evaluation
Different SNP chips3 K chip available
• Impute from 3 K to 50K use 50 K prediction formula
• Large potential for cows and screening of bullsHD Chip
• Potential for higher reliability (see presentation of A. Eggen)
Best use of GEBVs in breeding programsIncorporation of international genetics in domestic selection decisions(see presentation of J. Dürr about Interbull plans)
9. Juni 2010 Seite 19
Summary
Genomics leads to significant changes in genetic evaluation
Significant changes also in structure of breeding programs
Clear need for more collaborationBest done in the Interbull/ICAR framework
Breeding organisation and herd recording organisations will have to adapt their services and structure to the new opportunities